AWS says most large committed capacity is sold out through 2027

Jassy made the remarks in a video podcast with Andreessen Horowitz co-founder Ben Horowitz and general partner Erik Torenberg, reported by Reuters on October 23. He said the company keeps some capacity back for On Demand, EC2 and its Bedrock inference service, and that capacity for 2026 is already gone.

AWS sells capacity assurance through On-Demand Capacity Reservations and Capacity Blocks for ML, which let customers reserve instances in a named Availability Zone. InfoQ previously reported that AWS raised Capacity Blocks for ML rates by about 15% across every region in January.

Cloud analyst David Linthicum said the reporting overstated the position, since pre-reserved blocks for large model providers differ from capacity an enterprise can allocate day to day. A Reddit commenter described recent capacity shortages for regular EC2 instances, not just Spot.

Teams planning large GPU or compute needs should check reservation options and lead times rather than assume on-demand capacity will be available. Commenters also reported shortages for ordinary instance types, which can affect CI pipelines and production clusters.

The $220 billion figure in the report is total capital expenditure, not AI spending alone, so it should not be read as a measure of AI capacity.